Design of an Enhanced Deep Learning Model for Epileptic Seizure Assessment

Almost 65 million people worldwide have epilepsy, which is a highly prevalent neurological condition. More than 30% of the time, neither medication nor surgery, can cure those who have this disease. With therapeutic intervention, nevertheless, anticipating a seizure before it happens can help in averting it. According to previous information gathered, a period of aberrant brain anomaly called pre-ictal state begins prior to the start of seizure. There have been numerous attempts by researchers to forecast this pre-ictal state of a seizure, but it is still difficult to make a prediction with high sensitivity and specificity. The presented estimation model for seizure in the current study makes use of deep analytics approach. The preprocessing of scalp EEG signals, automatic feature uprooting, use of convolution neural model, and categorization with assistance from vector models are all included in this method. Using the suggested technique on 24 participants in the scalp EEG dataset of CHB-MIT, an average responsiveness and accuracy of 92.7% and 90.8%, respectively, were effectively achieved.